1. Field of the Invention
The present invention relates to a method, system, and article of manufacture for using device status information to takeover control of devices assigned to a node.
2. Description of the Related Art
Processing units in a system may share multiple adapters that enable communication with one or more storage systems. The processing units may communicate with the adapters via one or more networks. If one of the processing units fails, a surviving processing unit may take over the adapter assigned/owned by the failing processor. If the takeover occurred while an adapter owned by the surviving processing unit failed, then the takeover would allow the processing unit to maintain connectivity to an adapter and the connected storage system to maintain continued access to the storage systems accessible through the adapters.
The surviving processing unit needs to determine the status of the adapters assigned to the failed/failing processing unit to determine whether it may be taken over. In certain configurations, the surviving system may not be able to directly query the adapter assigned to the failing processing unit to determine the status because only the assigned/owner processing unit, i.e., the failing processing unit, can communicate with the adapter. In such case, the surviving processor may nonetheless try to takeover the adapter owned by the failing processing unit without querying the adapter to takeover to determine whether it is in fact available.
There is a need in the art for improved techniques to handle a takeover of a device or adapter assigned to a failed or failing node.